Research in Microbiology 153 (2002) 205212 www.elsevier.

com/locate/resmic

Mini-review

Molecular biology of cellulose production in bacteria

Ute Rmling 1
Department of Cell Biology and Immunology, Research group Clonal variability, GBF, 38124 Braunschweig, Germany Received 18 January 2002; accepted 8 March 2002

Abstract Cellulose biosynthesis has recently been established for a variety of bacteria of diverse origin at the phenotypic and genetic levels. Novel regulatory pathways, which involve the second messenger bis-(3 ,5 ) cyclic diguanylic acid and several proteins with the GGDEF domain, participate in the regulation of cellulose biosynthesis. The biological signicance of cellulose production in environmental, commensal and pathogenic bacteria is only punctually resolved. This review summarizes current knowledge on cellulose biosynthesis, its regulation and biological function. 2002 ditions scientiques et mdicales Elsevier SAS. All rights reserved.
Keywords: Cellulose biosynthesis; Acetobacter xylinus; Salmonella typhimurium; c-di-GMP

1. Introduction

2. Detection of cellulose production Although only consisting of the monosaccharide glucose in (1 4) -glycosidic bonds the polysaccharide cellulose is both a relatively simple and a complicated macromolecule. Several dozen (in some organisms up to 250) linear glucan chains are arranged in parallel and form highly regular intra- and interchain hydrogen bonds. In nature, the most frequently found crystalline structure is the metastable cellulose I allomorph. This particular tertiary structure actually denes the characteristics of cellulose that consist of water-insoluble crystalline microbrils inert to the treatment even with strong alkaline and acidic solutions [40]. This peculiarity in the behavior of the macromolecule is responsible for the fact that production of cellulose by bacteria is not detected using conventional methodology for the analysis of exopolysaccharide production. However, convenient primary screens for cellulose-producing organisms are provided by the characteristic binding behavior of dyes to the cellulose molecule such as Congo red or calcouor before detailed genetic and chemical analyses are established [2,6, 16,28]. Since cellulose production leads to the aggregation of bacteria, dissolution of cell clumps by the treatment with cellulase but not with other glucosidases such as amylase or proteinases is another hint that cellulose is produced [6,16, 44]. Cellulose production has been established for G. xylinus, Agrobacterium tumefaciens, Rhizobium leguminosarum bv. trifolii, Sarcina ventriculi and recently, for the enterobacteriaceae Salmonella spp., Escherichia coli, Klebsiella pneu-

Although cellulose production has been considered to be a domain of the plant kingdom, for a long time the model organism for the elucidation of basic features of cellulose biosynthesis has been the bacterium Gluconacetobacter xylinus (formerly called Acetobacter xylinus(m)). For example, efcient in vitro cellulose biosynthesis and the cloning of the cellulose biosynthesis operon in G. xylinus at the beginning of the nineties had been milestones in cellulose research [29,43]. To this end, based on the homology of conserved motives to the bacterial cellulose synthases, the rst plant cellulose synthase genes were recently identied [24]. Although the role of a bacterium as a model organism for cellulose biosynthesis is now superuous, the recent discovery of cellulose production (and the prediction of cellulose biosynthesis) in a wide variety of bacteria has opened up exciting perspectives for the elucidation of molecular mechanisms of cellulose biosynthesis and regulation, and for the role of cellulose in bacterial development and in the interaction of the bacterial cell with the environment and the host. In addition, cellulose production in bacteria is of potential economical interest.

E-mail address: ute.romling@mtc.ki.se (U. Rmling).

1 Present address: Microbiology and Tumorbiology Center (MTC),

Karolinska Institute, Box 280, S-17177 Stockholm, Sweden.

0923-2508/02/$ see front matter 2002 ditions scientiques et mdicales Elsevier SAS. All rights reserved. PII: S 0 9 2 3 - 2 5 0 8 ( 0 2 ) 0 1 3 1 6 - 5

206

U. Rmling / Research in Microbiology 153 (2002) 205212

Fig. 1. Conrmed and predicted bcs operons and identied regulatory genes of cellulose biosynthesis. Open reading frames (ORFs; not drawn to scale) that encode homologous genes (as determined experimentally or by BLAST search) have the same color. Alternative designations for homologous genes as assigned in databases are indicated within the respective arrow. Grey and white arrows indicate genes not proven to be involved in cellulose biosynthesis. A light blue box stands for a GGDEF domain. Sspp, Salmonella serotypes (serotype Typhimurium (NC_003197, AJ315770, AJ002301, AJ271071) ; serotype Typhi (NC_003198)); Ec, E. coli K12, O157:H7 EDL933, O157:H7 (NC_000913, AE005174, NC_002695); Pp, P. putida KT2440; Ps, P. syringae pv. tomato; Bf, Burkholderia sp. strain LB400; Rm, R. metallidurans CH34; G. xylinus, strain JCM7664 (I; AB015802); JCM7664 (II-A; AB015803); JCM7664 (II-B; AB015804); BPR2001 (AB010645); 1306 (M37202); ATCC53582 (X54676); AY201/ATCC23769 (U15957); ATCC23769 (M96060); At, A. tumefaciens C58 (NC_003063, L38609); Rl, R. leguminosarum bv. trifolii (AF121340, AF121341); Rs, Rhodobacter sphaeroides 2.4.1; Aa, A. aeolicus VF5 (NC_000918). Preliminary sequence data for P. putida and P. syringae were obtained from The Institute for Genomic Research website at http://www.tigr.org and for Burkholderia sp., R. metallidurans and R. sphaeroides at http://spider.jgi-psf.org/JGI_microbial/html.

moniae and several species of cyanobacteria [2,6,17,22,23, 30,44]. A cellulose synthase has been identied in all the above-mentioned organisms besides in the Gram-positive bacterium S. ventriculi. With the exception of the cyanobacteria, central structural genes required for cellulose biosynthesis, including the cellulose synthase, form an operon on the chromosome (Fig. 1). G. xylinus strains may have more than one cellulose biosynthesis operon, whereby cellulose biosynthesis is mediated only by one of the synthases in vivo under laboratory conditions [31]. Two genes are present in the cellulose biosynthesis operon throughout the species, the cellulose synthase and the bis-(3 ,5 ) cyclic diguanylic acid (c-di-GMP) binding protein. The cellulose synthase is the rst gene in the operon encoded by bcsA (bacterial cellulose synthesis), which is also named acsA (Acetobacter cellulose synthesis) or celA

(cellulose). Second comes the c-di-GMP binding protein encoded by bcsB (synonyms: acsB, celB). In G. xylinus the bcsA and bcsB genes are occasionally fused to a single open reading frame in type II cellulose synthases which shows the tight functional coupling of the two protein products for which spatial closeness has recently been demonstrated [12]. Otherwise, the present stage of knowledge suggests that there is variability in the location and possibly also in the requirement of additional genes for cellulose biosynthesis. BcsZ (synonyms CMC in G. xylinus and celC in A. tumefaciens and R. leguminosarum bv. trifolii) has been shown to encode a cellulase (family 8 glucosidase), which is required for cellulose synthesis. BcsZ present in all cellulose-producing species is encoded by the cellulose biosynthesis operons of enterobacterial species, A. tumefaciens and R. leguminosarum bv. trifolii, but located

U. Rmling / Research in Microbiology 153 (2002) 205212

207

outside yet adjacent to the cellulose biosynthesis operon in several G. xylinus strains [37]. BcsC required in vivo but not in vitro for cellulose biosynthesis is present in the enterobacterial species, the pseudomonads and G. xylinus. Whether other genes suspected of participating directly in cellulose biosynthesis such as bcsD (required in vivo, but not in vitro for cellulose biosynthesis [32,43]), ccp in G. xylinus [36] and celDE in A. tumefaciens and R. leguminosarum bv. trifolii [2,17] have functional homologues in the other cellulose-producing species remains to be shown. Search of the databases of nished and unnished bacterial genomes detected several bacterial species that contain homologues to the four genes of the enterobacterial bcsABZC operon on their chromosomes (Fig. 1). The genes of the bcs operons of two Pseudomonas species, Pseudomonas putida KT2440 and Pseudomonas syringae pv. tomato and of Burkholderia sp. strain LB400 are most closely related to the genes of the bcs operon of the enterobacteriaceae and show the same gene order. In Ralstonia metallidurans CH34 (formerly called Alcaligines/Ralstonia eutrophus) and Aquifex aeolicus VF5 a recombination event separated bcsA from bcsBZC, the rest of the operon. Burkholderia pseudomallei also contains highly conserved bcsABZC genes, but in the current state of the sequencing project it is not possible to determine whether they build up an operon. Based on the high homology of the genes it is expected that all those bacterial species mentioned above are capable of producing

cellulose. For several Pseudomonas spp. strains production of cellulose has already been described [1]. Consequently, cellulose, which had once been considered to be produced only by a few soil bacteria of the -branch of proteobacteria has become a common polysaccharide secreted by a variety of unrelated environmental, commensal and pathogenic bacteria.

3. The cellulose synthase BcsA The cellulose synthase BcsA is between 723 to 888 amino acids (aa) long and is the most conserved gene of the bcs operon among the species. Although the N- and C-terminal part of the protein is less well conserved the homology is not restricted to the frequently analyzed D,D,D35Q(R,Q)XRW motif, which spans domains A and B (Fig. 2). The D,D,D35Q(R,Q)XRW motif is actually characteristic of the whole group of processive -glycosyltransferases which include, besides cellulose synthase, among others chitin synthase and curdlan synthase. In bacterial cellulose synthases, conserved aa suggest calling the D,D,D35Q(R,Q) XRW motif the D3 D2 D35QRXRWA motif (Fig. 2). When conrmed and predicted cellulose synthases were compared a region of high homology ( 70% aa similarity) comprises approximately 350 aa. The region of high homology contains the D3 D2 D35QRXRWA motif and ve additional

Fig. 2. Consensus sequence of bacterial cellulose synthase, catalytic subunit BcsA. Experimentally conrmed and selected predicted BcsA proteins were used. Multiple alignment of the amino acid sequences was performed by PileUp (GCG package version 9, University of Wisconsin) using standard parameters. The consensus sequence was drawn using the SeqLogo program [33] in the WWW-based implementation (http://www.bio.cam.ac.us/seqlogo). Only aa that are more than 80% conserved are shown with BcsA from S. typhimurium as frame. The height of each letter reects the degree of conservation. The solid blue bar below the alignment indicates the regions of homology to processive glycosyltransferase domain A and domain B, respectively. Domain A overlaps with the glycosyltransferase 2 consensus sequences [4]. The solid red bar below the alignment indicates the region of homology to eukaryotic cellulose synthases with characteristic motives U1 to U4 [24]. The solid green bar below the alignment marks regions where the completely conserved amino acids of the cellulose synthases deviate from the sequence of the closest homologue, curdlan synthase from Agrobacterium spp., a (1 3) -glucan synthase [38]. Protein sequences used: G. xylinus: BAA31463.1, P19449, BAA77585.1, P21877, BAA77593.1; S. typhimurium: CAC44015.1; E. coli: P37653; A. tumefaciens: NP_357298.1; R. leguminosarum bv. trifolii: AAC41436.1. P. putida and P. syringae pv. tomato BcsA proteins according to sequence data from http://www.tigr.org, R. metallidurans, Burkholderia sp. strain LB400 and R. sphaeroides 2.4.1 BcsA proteins from http://spider.jgi-psf.org/JGI_microbial/html.

208

U. Rmling / Research in Microbiology 153 (2002) 205212

highly conserved sequence stretches which cluster around positively or negatively charged aa and contain the majority of invariable aa, namely the PVDPYE, HAKAGN(L,I)N, QTP, FFCGS and RFLPL motives (Fig. 2). Most closely related to the bacterial cellulose synthases is the curdlan synthase, a (1 3) -glucan synthase, from Agrobacterium spp. [38] which contains, besides the D,D,D35Q(R,Q)XRW motif, an identical HAKAGN(L,I)N motif, slightly altered PVDPYE and QTP motives, but almost completely lacks the FFCGS and RFLPL motives. This nding suggests that the FFCGS and RFLPL motives could be mainly responsible for the determination of the (1 4) specicity of the -glucan bond. Divergence among the cellulose synthases is highest in an aa stretch between D3 and the HAKAGN motif where the CelA and BcsA proteins contain a 34 and 15 aa long insertion, respectively. The cellulose synthase is located in the cytoplasmic membrane with 8 to 10 predicted transmembrane domains. BcsA is considered to be the catalytic subunit for cellulose biosynthesis. Besides the high homology to other processive glycosyltransferases, BcsA has been experimentally shown to bind the substrate UDP-glucose [13].

In G. xylinus the cellulose-synthesizing complex is a transmembrane complex over the cytoplasmic and outer membrane whereby the cellulose synthase (BcsA) and the c-di-GMP binding protein (BcsB) are considered to be localized in the cytoplasmic membrane [12]. In cells actively producing cellulose approximately 50 cellulose-synthesizing multienzyme complexes are organized in a single row along the longitudinal axis of the bacterial rod whereby each complex secretes approx. 12 to 25 glucan chains which assemble into larger microbrils at the site of synthesis. This so-called linear terminal complex can be visualized by electron microscopy using freeze fracture as 35 pores in the outer membrane or as pits when the outer leaet is fractured away [12]. At the molecular level the mechanism of cellulose biosynthesis has not been resolved. In A. tumefaciens participation of two different lipid intermediates in cellulose biosynthesis has been postulated whereby the initial glucose-lipid derivative is formed by the celDE gene product [18]. There is no hint of lipid intermediates in cellulose biosynthesis nor in the occurrence of celDE homologues in G. xylinus.

6. c-di-GMP, activator of cellulose biosynthesis 4. The BcsB protein The BcsB protein that was indirectly inferred to bind c-di-GMP [19] is less well conserved among the species. However, direct comparisons of the BcsB proteins with CelB from A. tumefaciens and R. leguminosarum bv. trifolii revealed signicant homology ( 40% similarity) over the entire length of the proteins with several invariable residues (Fig. 3). An alanine/proline rich domain is located at the N-terminus of all proteins except A. aeolicus. One transmembrane domain located at the C-terminus has been predicted by various algorithms for all BcsB proteins. In G. xylinus c-di-GMP has been identied as an activator of cellulose biosynthesis [29]. The free c-di-GMP in the cell is considered to allosterically activate the cellulose synthase BcsA. However, 90% of the cellular c-di-GMP is reversibly bound by the c-di-GMP binding protein BcsB, a membrane protein that is structurally associated with the cellulose synthase [12,19,43]. It is believed that the spatial proximity is necessary to direct c-di-GMP released from BcsB towards the cellulose synthase. The equilibrium between bound and free c-di-GMP is modulated by the intracellular potassium concentration [42]. The level of free c-di-GMP is regulated by the opposing action of two enzymes, diguanylate cyclase (DGC) that cycles two molecules of GTP under the release of two molecules of PPi , and phosphodiesterase A (PDEA) that degrades c-di-GMP to the inactive pGpG. G. xylinus has three distinct operons each containing a PDEA/DGC pair (Fig. 1), which contribute at different levels to the c-di-GMP turnover [39] indicating that cellulose biosynthesis underlies various control mechanisms in G. xylinus. At the N-terminus DGC as well as PDEA contain sensory domains for environmental signals [39, (Fig. 4)]. The

5. Biosynthesis of cellulose Although elucidation of the structure and function of the cellulose-synthesizing complex will be useful for the general understanding of membrane complexes that traverse both the inner and the outer membrane as well as for the directed manipulation of cellulose production, surprisingly little is known about the function and localization of the proteins participating in cellulose biosynthesis.

Fig. 3. Signature of completely conserved residues in BcsB/CelB proteins with BcsB from S. typhimurium (CAC44016.1) as an example. ; 100 aa.

U. Rmling / Research in Microbiology 153 (2002) 205212

209

Fig. 4. Comparison of the domain structures of proteins involved in the regulation of cellulose biosynthesis. PDEA1 and DGC1 from G. xylinus (AF052517); CelR2 from R. leguminosarum bv. trifolii (AF121341): AdrA from S. typhimurium and E. coli (AJ271071). Heme-PAS is an oxygen-sensing domain similar to a domain in FixL from rhizobia, whereas avin-PAS has homology to a redox-sensing domain in NifL from K. pneumoniae [39]; CheY indicates a domain homologous to the receiver domain of two-component systems; the GGDEF and EAL domains are described in the text. Shaded boxes indicate putative membrane domains.

activity of PDEA has been shown to be negatively regulated by oxygen sensed by a heme-containing PAS domain [5] DGC has been proposed to harbor a avin-containing PAS domain, which senses the redox status of the cell. DGC and PDEA are homologous multidomain proteins that have a GGDEF (also called DUF1) and EAL (synonym: DUF2) domain in common (Fig. 4). Both domains are widely represented in bacteria and may occur alone, together or with well-established domains for signal transduction [8]. Recently, the detection of sequence similarities between the GGDEF domain and eukaryotic adenylate cyclases predicted a nucleotide cyclase activity as a possible function for the GGDEF domain [25]. In that line, unrelated proteins of different bacteria that only had the GGDEF domain in common were able to induce cellulose biosynthesis in a cellulose-decient R. leguminosarum bv. trifolii wild-type strain when plasmid-expressed [3]. However, there exists no truly convincing explanation as to why certain bacterial species like E. coli and S. typhimurium have about 20 proteins with the GGDEF-based nucleotide cyclase activity all producing the soluble activator molecule c-di-GMP. The function of the EAL domain still remains unknown. Although there is evidence, it is not yet clear whether c-di-GMP acts in general as an activator of cellulose biosynthesis in bacteria. Besides G. xylinus, the occurrence of c-di-GMP and activation of cellulose biosynthesis by c-diGMP has only been shown for A. tumefaciens [1]. The discovery of DOS, a protein encoded by E. coli and S. typhimurium that is highly homologous to PDEA from G. xylinus [7], is a hint of the occurrence of c-di-GMP in those organisms and at least partially overlapping signal transduction pathways leading to cellulose biosynthesis. Strikingly, by genetic analysis, individual proteins with GGDEF domains have been identied as participating in the positive regulation of cellulose biosynthesis in R. leguminosarum bv. trifolii and S. typhimurium [2,27]. The GGDEF domain is the only homologous domain among the proteins (Fig. 4). Otherwise, CelR2, which controls cellulose biosynthesis in R. leguminosarum bv. trifolii, is homologous to PleD, a response regulator of cell differentiation in Caulobacter crescentus [10]. On the other hand, AgfDregulated protein (AdrA), a regulator of cellulose biosynthe-

sis in S. typhimurium, has four transmembrane N-terminal domains of the GGDEF domain. Yet too little is known about the regulation of cellulose biosynthesis to predict distinct roles for these proteins at the molecular level. Based on the few phenotypes that are known for genes encoding GGDEFdomain-containing proteins it can be concluded that GGDEF domains play a role in bacterial morphogenesis.

7. Regulation of cellulose biosynthesis in S. typhimurium In S. typhimurium a regulatory cascade that leads to cellulose biosynthesis has been elucidated using a strain that shows semi-constitutive cellulose expression [27,44]. AgfD, a response regulator of the LuxR-family that was initially detected as regulating the biosynthesis of thin aggregative mbriae [26], was shown to transcriptionally regulate adrA under a variety of environmental conditions [27]. The expression of AgfD itself is mainly restricted to the stationary phase of growth and subject to regulation by global regulatory proteins and a variety of environmental conditions [26]. Expression of AdrA in turn activates cellulose biosynthesis even when agfD is deleted, while adrA is plasmidexpressed [44]. Since it has been shown that transcription of bcsA and bcsC is not dependent on AgfD and therefore AdrA, activation of cellulose biosynthesis by AdrA must occur at a posttranscriptional level [44]. Stabilization or activation of the Bcs proteins by protein-protein interactions with AdrA or production of c-di-GMP by the GGDEF domain of AdrA could trigger cellulose biosynthesis. Since the above described genes involved in cellulose biosynthesis are highly conserved among S. typhimurium, other Salmonella serotypes and E. coli, it is expected that a similar regulatory cascade leading to cellulose biosynthesis works in the respective organisms.

8. Biological signicance of cellulose production in bacteria Most knowledge of the biological role of cellulose biosynthesis has been gathered on the interaction of soil

210

U. Rmling / Research in Microbiology 153 (2002) 205212

bacteria of the family Rhizobiaceae, Rhizobium spp. and A. tumefaciens, respectively, with plants. Rhizobium spp. are plant symbionists that x nitrogen when living inside root nodule cells of leguminous host plants. After the initial attachment of the individual bacterial cells to plant root hairs in culture, cellulose biosynthesis is required for the second step in adherence, rm adherence plus aggregation of bacteria at the root hair tip (cap formation) [36]. However, cap formation is not a prerequisite for nodulation, the next step in the interaction of the bacteria with the plant. The physiological role of cap formation, although performed by various Rhizobium spp. [35], is not resolved. A contribution of cellulose to more effective nodulation has been proposed [22], but never convincingly proven. A similar role for cellulose in sequential attachment to carrot tissue culture cells has been shown for A. tumefaciens the causative agent of crown gall tumours on dicotyledonous plants [14]. Mimicking environmental conditions colonization of roots in soil was signicantly reduced by cellulose mutants of A. tumefaciens [15]; however, the survival rate of the wildtype and the cellulose mutants in soil has not been determined. Anchoring of the bacterial cells to the plant tissue, which leads to a survival advantage under natural conditions, might be a function of cellulose biosynthesis in Rhizobiaceae. A similar explanation concerning the function of cellulose might hold for G. xylinus, considering that the natural environment of G. xylinus is the surface of fruits, vegetables and decaying material [34]. Although the most popular argument for cellulose biosynthesis in G. xylinus is the maintenance of the organism in an aerobic environment in liquids [11], this view might be biased by observations from a laboratory perspective. The effect of cellulose on the virulence of A. tumefaciens in a leaf infection model was not consistent, since some of the cellulose-minus mutants were not affected in virulence, whereas others showed reduced virulence [20]. This nding suggests that some mutations occurred in genes that regulate cellulose biosynthesis but also affect other cellular functions. Another phenotype that has been observed with cellulose pellicles is the protection of cells from the hazardous effect of UV radiation [34]. Although cellulose biosynthesis is common among several species of the Enterobacteriaceae the biological function of cellulose in these organisms is not clear. Cellulose is produced by the majority of S. typhimurium and S. enteritidis isolates from disease origin, but not by S. typhimurium isolates from doves and other Salmonella spp. with a narrow host range which cause invasive disease (unpublished observations). This fact indicates that there is a strong selective pressure against cellulose production under certain conditions in the host. I hypothesize that cellulose biosynthesis in Enterobacteriaceae once played a role in the bacterial life cycle outside the animal host in biolm formation, cell-cell interaction and/or persistence of the organisms. In that environment species of Enterobacteriaceae might have acquired the cellulose biosynthesis operon bcsABZC and adrA, the ac-

tivator of cellulose biosynthesis, from Pseudomonas spp. or related soil organisms, as concluded from the high homology of the respective operons on the nucleotide and amino acid level [44]. Whether a critical (positive) role for cellulose biosynthesis in bacterial-host interaction emerged later in evolution remains to be determined. Cellulose biosynthesis in Salmonella spp. and E. coli occurs concomitantly with the production of thin aggregative mbriae (AGF), the second component of the extracellular matrix of a multicellular morphotype [27]. Each of these substances mediates a specic type of cell-cell as well as cell-surface interaction [27]. While thin aggregative mbriae form rigid, but fragile interconnections between cells, cellulose connects the cells through elastic, but stable bonds. One developmental characteristic of the multicellular morphotype is biolm formation on abiotic surfaces where cells producing cellulose and thin aggregative mbriae form distinct adherence patterns. In a steady state model, cells which express AGF mbriae form a hard-to-remove biolm that starts below the air-liquid interface, while celluloseexpressing cells form a loosely adherent biolm at the air-liquid interface [27,44]. When expressed together in a colony on plates the two substances form a highly inert, hydrophobic extracellular matrix around the cells which enables the bacteria to act like a multicellular organism and not as individual cells as detected by various types of phenotypic assays [27]. This developmental behavior of Salmonella and E. coli strains has been named the rdar (red, dry and rough) morphotype inspired by the characteristic phenotype of the colony exhibited on Congo red containing agar plates [26].

9. Impact of bacterial cellulose production in medical settings and industrial applications Besides its role in the natural environment cellulose biosynthesis of bacterial organisms has its impact in medical settings. Enterobacteriaceae, in particular E. coli, frequently cause nosocomial infections such as sepsis, biliary tract infections and catheter-related cystitis that are caused by biolm-forming isolates. Since the major components of the extracellular matrix of biolm forming E. coli have been identied, more rational approaches to prevent adhesion to catheter material can be designed. Plant-derived cellulose is used in high quantities as a starting material in various industrial branches. Since bacterial cellulose is of high purity and displays special physico-chemical characteristics, it has found numerous applications in the paper and food industry (as acoustic membrane and food texture) and in the medical eld (as an articial skin and blood vessel substitute) [9,41]. However, the industrial production of bacterial cellulose is yet fairly inefcient, although strain selection, genetic manipulation and process optimisation have been used to enhance productivity of cellulose [21,41].

U. Rmling / Research in Microbiology 153 (2002) 205212

211

10. Perspectives For some time cellulose biosynthesis has been receiving attention in basic as well as applied research. Until recently, basic questions about cellulose biosynthesis had been answered in the model organism G. xylinus. With the discovery of cellulose biosynthesis in well-characterized organisms such as E. coli and S. typhimurium cellulose biosynthesis can be elucidated using all the well-established molecular tools and sequence information available for these organisms. A better understanding of the molecular mechanisms of cellulose biosynthesis and regulation will also help to develop strategies for the eradication of biolm-forming bacteria and the optimization of cellulose production for industrial applications.

Acknowledgements I would like to thank the members of my group and collaborators who contributed to the fundamentals of this review; in particular, X. Zogaj and M. Nimtz. W. Rabsch and H. Tschpe are gratefully acknowledged for collaboration. This work was supported in part by the Bundesministerium fr Forschung und Technologie (BMFT) program Infektionsbiologie and by the Deutsche Forschungsgemeinschaft (Ro2023/3-1). Release of sequence data prior to publication from the Sanger Institute, TIGR, JGI and the University of Washington is gratefully acknowledged.

References
[1] D. Amikam, M. Benziman, Cyclic diguanylic acid and cellulose synthesis in Agrobacterium tumefaciens, J. Bacteriol. 171 (1989) 66496655. [2] N. Ausmees, H. Jonsson, S. Hglund, H. Ljunggren, M. Lindberg, Structural and putative regulatory genes involved in cellulose synthesis in Rhizobium leguminosarum bv. trifolii, Microbiology 145 (1999) 12531262. [3] N. Ausmees, R. Mayer, H. Weinhouse, G. Volman, D. Amikam, M. Benziman, M. Lindberg, Genetic data indicate that proteins containing the GGDEF domain possess diguanylate cyclase activity, FEMS Microbiol. Lett. 204 (2001) 163167. [4] J.A. Campbell, G.J. Davies, V. Bulone, B. Henrissat, A classication of nucleotide-diphospho-sugar glycosyltransferases based on amino acid sequence similarities, Biochem. J. 326 (1997) 929939. [5] A.L. Chang, J.R. Tuckerman, G. Gonzalez, R. Mayer, H. Weinhouse, G. Volman, D. Amikam, M. Benziman, M.A. Gilles-Gonzalez, Phosphodiesterase A1, a regulator of cellulose synthesis in Acetobacter xylinum, is a heme-based sensor, Biochemistry 40 (2001) 34203426. [6] M.H. Deinema, L.P. Zevenhuizen, Formation of cellulose brils by Gram-negative bacteria and their role in bacterial occulation, Arch. Mikrobiol. 78 (1971) 4251. [7] V.M. Delgado-Nixon, G. Gonzalez, M.A. Gilles-Gonzalez, Dos, a heme-binding PAS protein from Escherichia coli, is a direct oxygen sensor, Biochemistry 39 (2000) 26852691. [8] M.Y. Galperin, A.N. Nikolskaya, E.V. Koonin, Novel domains of the prokaryotic two-component signal transduction systems, FEMS Microbiol. Lett. 203 (2001) 1121.

[9] U. Geyer, T. Heinze, A. Stein, D. Klemm, S. Marsch, D. Schumann, H.P. Schmauder, Formation, derivatization and applications of bacterial cellulose, Int. J. Biol. Macromol. 16 (1994) 343347. [10] G.B. Hecht, A. Newton, Identication of a novel response regulator required for the swarmer-to-stalked-cell transition in Caulobacter crescentus, J. Bacteriol. 177 (1995) 62236229. [11] S. Hestrin, M. Schramm, Synthesis of cellulose by Acetobacter xylinum, Biochemistry 58 (1954) 345352. [12] S. Kimura, H.P. Chen, I.M. Saxena, R.M. Brown Jr., T. Itoh, Localization of c-di-GMP-binding protein with the linear terminal complexes of Acetobacter xylinum, J. Bacteriol. 183 (2001) 56685674. [13] F.C. Lin, R.M. Brown Jr., R.R. Drake Jr., B.E. Haley, Identication of the uridine 5 -diphosphoglucose (UDP-Glc) binding subunit of cellulose synthase in Acetobacter xylinum using the photoafnity probe 5 -azido-UDP-Glc, J. Biol. Chem. 265 (1990) 47824784. [14] A.G. Matthysse, Role of bacterial cellulose brils in Agrobacterium tumefaciens infection, J. Bacteriol. 154 (1983) 906915. [15] A.G. Matthysse, S. McMahan, Root colonization by Agrobacterium tumefaciens is reduced in cel, attB, attD, and attR mutants, Appl. Environ. Microbiol. 64 (1998) 23412345. [16] A.G. Matthysse, K.V. Holmes, R.H. Gurlitz, Elaboration of cellulose brils by Agrobacterium tumefaciens during attachment to carrot cells, J. Bacteriol. 145 (1981) 583595. [17] A.G. Matthysse, S. White, R. Lightfoot, Genes required for cellulose synthesis in Agrobacterium tumefaciens, J. Bacteriol. 177 (1995) 10691075. [18] A.G. Matthysse, D.L. Thomas, A.R. White, Mechanism of cellulose synthesis in Agrobacterium tumefaciens, J. Bacteriol. 177 (1995) 10761081. [19] R. Mayer, P. Ross, H. Weinhouse, D. Amikam, G. Volman, P. Ohana, R.D. Calhoon, H.C. Wong, A.W. Emerick, M. Benziman, Polypeptide composition of bacterial cyclic diguanylic acid-dependent cellulose synthase and the occurrence of immunologically crossreacting proteins in higher plants, Proc. Natl. Acad. Sci. USA 88 (1991) 5472 5476. [20] S.L. Minnemeyer, R. Lightfoot, A.G. Matthysse, A semiquantitative bioassay for relative virulence of Agrobacterium tumefaciens strains on Bryophyllum daigremontiana, J. Bacteriol. 173 (1991) 77237724. [21] T. Nakai, N. Tonouchi, T. Konishi, Y. Kojima, T. Tsuchida, F. Yoshinaga, F. Sakai, T. Hayashi, Enhancement of cellulose production by expression of sucrose synthase in Acetobacter xylinum, Proc. Natl. Acad. Sci. USA 96 (1999) 1418. [22] C. Napoli, F. Dazzo, D. Hubbell, Production of cellulose microbrils by Rhizobium, Appl. Microbiol. 30 (1975) 123131. [23] D.R. Nobles, D.K. Romanovicz, R.M. Brown Jr., Cellulose in cyanobacteria. Origin of vascular plant cellulose synthase?, Plant Physiol. 127 (2001) 529542. [24] J.R. Pear, Y. Kawagoe, W.E. Schreckengost, D.P. Delmer, D.M. Stalker, Higher plants contain homologs of the bacterial celA genes encoding the catalytic subunit of cellulose synthase, Proc. Natl. Acad. Sci. USA 93 (1996) 1263712642. [25] J. Pei, N.V. Grishin, GGDEF domain is homologous to adenylyl cyclase, Proteins 42 (2001) 210216. [26] U. Rmling, W.D. Sierralta, K. Eriksson, S. Normark, Multicellular and aggregative behavior of Salmonella typhimurium strains is controlled by mutations in the agfD promoter, Mol. Microbiol. 28 (1998) 249264. [27] U. Rmling, M. Rohde, A. Olsen, S. Normark, J. Reinkster, AgfD, the checkpoint of multicellular and aggregative behavior in Salmonella typhimurium regulates at least two independent pathways, Mol. Microbiol. 36 (2000) 1023. [28] U. Rmling, Genetic and phenotypic analysis of multicellular behavior in Salmonella typhimurium, in: R.J. Doyle (Ed.), Methods Enzymol., Academic Press, San Diego, pp. 4859. [29] P. Ross, H. Weinhouse, Y. Aloni, D. Michaeli, P. Weinberger-Ohana, R. Mayer, S. Braun, E. de Vroom, G.A. van der Marel, J.H. van Boom, M. Benziman, Regulation of cellulose synthesis in Acetobacter xylinum by cyclic diguanylic acid, Nature 325 (1987) 279281.

212

U. Rmling / Research in Microbiology 153 (2002) 205212

[30] P. Ross, R. Mayer, M. Benziman, Cellulose biosynthesis and function in bacteria, Microbiol. Rev. 55 (1991) 3558. [31] I.M. Saxena, R.M. Brown Jr., Identication of a second cellulose synthase gene (acsAII ) in Acetobacter xylinum, J. Bacteriol. 177 (1995) 52765283. [32] I.M. Saxena, K. Kudlicka, K. Okuda, R.M. Brown Jr., Characterization of genes in the cellulose-synthesizing operon (acs operon) of Acetobacter xylinum: Implications for cellulose crystallization, J. Bacteriol. 176 (1994) 57355752. [33] T.D. Schneider, R.M. Stephens, Sequence logos: A new way to display consensus sequences, Nucleic Acids Res. 18 (1990) 60976100. [34] W. Scott Williams, R.E. Cannon, Alternative environmental roles for cellulose produced by Acetobacter xylinum, Appl. Environ. Microbiol. 55 (1989) 24482452. [35] G. Smit, J.W. Kijne, B.J. Lugtenberg, Correlation between extracellular brils and attachment of Rhizobium leguminosarum to pea root hair tips, J. Bacteriol. 168 (1986) 821827. [36] G. Smit, S. Swart, B.J. Lugtenberg, J.W. Kijne, Molecular mechanisms of attachment of Rhizobium bacteria to plant roots, Mol. Microbiol. 6 (1992) 28972903. [37] R. Standal, T.G. Iversen, D.H. Coucheron, E. Fjaervik, J.M. Blatny, S. Valla, A new gene required for cellulose production and a gene encoding cellulolytic activity in Acetobacter xylinum are colocalized with the bcs operon, J. Bacteriol. 176 (1994) 665672. [38] S.J. Stasinopoulos, P.R. Fisher, B.A. Stone, V.A. Stanisich, Detection of two loci involved in (1 3)-beta-glucan (curdlan) biosynthesis by

[39]

[40] [41]

[42]

[43]

[44]

Agrobacterium sp. ATCC31749, and comparative sequence analysis of the putative curdlan synthase gene, Glycobiology 9 (1999) 3141. R. Tal, H.C. Wong, R. Calhoon, D. Gelfand, A.L. Fear, G. Volman, R. Mayer, P. Ross, D. Amikam, H. Weinhouse, A. Cohen, S. Sapir, P. Ohana, M. Benziman, Three cdg operons control cellular turnover of cyclic di-GMP in Acetobacter xylinum: Genetic organization and occurrence of conserved domains in isoenzymes, J. Bacteriol. 180 (1998) 44164425. D.M. Updegraff, Semimicro determination of cellulose in biological materials, Anal. Biochem. 32 (1969) 420424. E.J. Vandamme, S. De Baets, A. Vanvaelen, K. Joris, P. De Wulf, Improved production of bacterial cellulose and its application potential, Polym. Degrad. Stab. 59 (1998) 9399. H. Weinhouse, S. Sapir, D. Amikam, Y. Shilo, G. Volman, P. Ohana, M. Benziman, c-di-GMP-binding protein, a new factor regulating cellulose synthesis in Acetobacter xylinum, FEBS Lett 416 (1997) 207211. H.C. Wong, A.L. Fear, R.D. Calhoon, G.H. Eichinger, R. Mayer, D. Amikam, M. Benziman, D.H. Gelfand, J.H. Meade, A.W. Emerick, R. Bruner, A. Ben-Bassat, R. Tal, Genetic organization of the cellulose synthase operon in Acetobacter xylinum, Proc. Natl. Acad. Sci. USA 87 (1990) 81308134. X. Zogaj, M. Nimtz, M. Rohde, W. Bokranz, U. Rmling, The multicellular morphotypes of Salmonella typhimurium and Escherichia coli produce cellulose as the second component of the extracellular matrix, Mol. Microbiol. 39 (2001) 14521463.